The purpose of the proposed research is to develop computer methods to predict the three-dimensional structures of protein complexes, given the three-dimensional structures of the individual protein components. A special goal of this work is to take into account the fact that there is some conformational variation at protein surfaces, due to the motions of the atoms at the surface. Each protein is represented by a cubical lattice of occupancies or densities, which are numbers between 0.0 and 1.0 describing how much of the time that cube is occupied by the protein's atoms. For cubes in the interior of the protein, the densities will be 1.0. The shapes of these soft or fuzzy surfaces will be analyzed, and complementary shapes will be docked together in the computer. Protein-protein docking predictions should be useful in (1) the assembly of viral coat protein monomers into capsids, (2) the genetic disease sickle cell anemia, where there is unwanted polymerization of hemoglobin subunits, (3) understanding antibody-antigen recognition, (4) understanding the mechanisms of protein toxins, such as diphtheria and cholera toxins, and (5) designing protein drugs. PROPOSED COMMERCIAL APPLICATION: The computer software developed during this project will be licensed to pharmaceutical and biotechnology companies for use in designing proteins to be used as pharmaceuticals. It will also be licensed, for a more modest fee, to universities and research institutes.